(CXCR4) Antagonists - ACS Publications - American Chemical Society

Aug 29, 2016 - We previously reported the discovery of a CXCL12-mimetic cyclic peptide (2) as a selective CXCR4 antagonist showing promising in vitro ...
1 downloads 0 Views 6MB Size
Article pubs.acs.org/jmc

Exploring the N‑Terminal Region of C‑X‑C Motif Chemokine 12 (CXCL12): Identification of Plasma-Stable Cyclic Peptides As Novel, Potent C‑X‑C Chemokine Receptor Type 4 (CXCR4) Antagonists Salvatore Di Maro,† Anna Maria Trotta,‡ Diego Brancaccio,§,∥ Francesco Saverio Di Leva,§ Valeria La Pietra,§ Caterina Ieranò,‡ Maria Napolitano,‡ Luigi Portella,‡ Crescenzo D’Alterio,‡ Rosa Anna Siciliano,⊥ Deborah Sementa,§ Stefano Tomassi,§ Alfonso Carotenuto,§ Ettore Novellino,§ Stefania Scala,*,‡ and Luciana Marinelli*,§ †

DiSTABiF, Second University of Naples, Via Vivaldi 43, 81100 Caserta, Italy Genomica Funzionale, Istituto Nazionale per lo Studio e la Cura dei Tumori, Fondazione “Giovanni Pascale”, IRCCS-ITALY, Via M. Semmola, 80131 Naples, Italy § Dipartimento di Farmacia, Università degli Studi di Napoli “Federico II”, via D. Montesano 49, 80131 Naples, Italy ∥ Laboratory of Food Chemistry, Dipartimento di Agraria (QuaSic.A.Tec.), Università Mediterranea di Reggio Calabria, Reggio Calabria, loc. Feo di Vito, 89122 Reggio Calabria, Italy ⊥ Istituto di Scienze dell’Alimentazione, CNR, Via Roma 64, 83100 Avellino, Italy ‡

S Supporting Information *

ABSTRACT: We previously reported the discovery of a CXCL12-mimetic cyclic peptide (2) as a selective CXCR4 antagonist showing promising in vitro and in vivo anticancer activity. However, further development of this peptide was hampered by its degradation in biological fluids as well as by its low micromolar affinity for the receptor. Herein, extensive chemical modifications led to the development of a new analogue (10) with enhanced potency, specificity, and plasma stability. A combined approach of Ala-amino acid scan, NMR, and molecular modeling unraveled the reasons behind the improved binding properties of 10 vs 2. Biological investigations on leukemia (CEM) and colon (HT29 and HCT116) cancer cell lines showed that 10 is able to impair CXCL12-mediated cell migration, ERKphosphorylation, and CXCR4 internalization. These outcomes might pave the way for the future preclinical development of 10 in CXCR4 overexpressing leukemia and colon cancer.



INTRODUCTION Chemokines constitute a family of small cytokines that regulate immunological and inflammatory processes, such as cell chemotaxis, hematopoiesis, and angiogenesis through the interaction with their G protein-coupled receptors (GPCRs).1 Activation of chemokine receptors by their physiological ligands elicits conformational changes that trigger a signaling cascade involving G protein binding, activation of kinases, and Ca2+ mobilization from cellular stores, which in turn induce cytoskeletal rearrangement and chemotaxis toward cell attractant molecules.2,3 The chemokine CXCL12, a 68-residue polypeptide, also known as stromal cell-derived factor-1 (SDF-1α), belongs to the CXC chemokines subfamily, which feature two conserved cysteine residues at specific positions in the N-terminal region.4 CXCL12 is the endogenous ligand of CXCR4, which is one of 19 known human chemokine receptors, and it is primarily coupled with Gi proteins to activate divergent signaling © 2016 American Chemical Society

pathways that are involved in multiple important physiological functions such as the development of the hematopoietic, nervous, gastrointestinal, and immune systems.5 CXCL12/ CXCR4 axis has been implicated in a wide range of diseases such as infections,6 inflammations,7 and cancer.8 Thus, the inhibition of the CXCL12/CXCR4 pathway has emerged as a novel anticancer strategy, widely supported by a large number of in vitro and in vivo studies.9 In particular, CXCR4 has been shown to play a prognostic role in primary colorectal cancer, alone or in association with the vascular endothelial growth factor (VEGF).10 In fact, CXCR4 can promote angiogenesis in vitro by modulating the secretion of VEGF.11,12 The pursuit of CXCR4 antagonists suitable for anticancer intervention culminated in the identification of plerixafor,12 previously known as AMD3100 (1,1′-[1,4-phenylenebisReceived: May 13, 2016 Published: August 29, 2016 8369

DOI: 10.1021/acs.jmedchem.6b00695 J. Med. Chem. 2016, 59, 8369−8380

Journal of Medicinal Chemistry



Article

RESULTS AND DISCUSSION Design. The peptide (H-Arg-Ala-[Cys-Arg-Phe-Phe-Cys]COOH) (2), previously identified by some of us, inhibits the association between CXCR4 and its ligand CXCL12 at a low micromolar level (IC50 = 6.2 μM) and in turn impairs CXCR4 function using in vitro and in vivo systems.19 However, as already stated in the Introduction, it displays low stability in biological fluids, as in human serum, 2 starts to be metabolized by 30% after only 30 min and is fully converted into the cyclic pentapeptide cyclo[Cys-Arg-Phe-Phe-Cys] within 2 h. Interestingly, the inactivation of 2 proceeds through the sequential cleavage of the exocyclic Arg1-Ala2 dipeptide at the N-terminal region. Additionally, the IC50 of 2 (Table 1) needs to be

(methylene)]bis[1,4,8,11-tetraazacyclotetradecane]) (Chart 1). In preclinical models, plerixafor (1) has shown considerable Chart 1. Chemical Structure of 1 (Plerixafor)

antimetastatic properties, thus, offering the first proof of concept of CXCR4’s importance as anticancer target. 1 has been approved from FDA for hematopoietic stem cell mobilization in patients with non-Hodgkin’s lymphoma and multiple myeloma undergoing autologous bone marrow transplantation and resistant to conventional protocol for mobilization.13 Recently, new alternative CXCR4 antagonists have been developed and subjected to preclinical and clinical studies.14 Among them, the (S)-N1-((1H-benzo[d]imidazol-2yl)methyl)-N1-(5,6,7,8-tetrahydroquinolin-8-yl)butane-1,4-diamine (AMD11070), which is endowed with oral bioavailability, had a trial suspended due to nonclinical reports of hepatotoxicity and histologic findings.15 Alternatively, some peptide antagonists, such as the peptides 4F-benzoyl-Arg-ArgNal-Cys-Tyr-Cit-Lys- D -Lys-Pro-Tyr-Arg-Cit-Cys-Arg-NH 2 (BKT140)16 and H-Lys-Gly-Val-Ser-Leu-Ser-Tyr-Arg-Lys(LysGly-Val-Ser-Leu-Ser-Tyr-Arg*)-NH2 (CTCE-9908),17 also moved to clinical studies, but their application appeared to be partially hampered by in vivo instability and other pharmacokinetic issues.16,17 Cyclo[Phe-Tyr-Lys(iPr)-D-Arg-2-Nal-Gly-DGlu]-Lys(iPr)-NH2 (LY2510924),18 a CXCR4 cyclic antagonist peptide currently engaged in two phase 2 studies in small cell lung cancer and renal cancer, shows CD34+ cell mobilization at doses ≥2.5 mg/day and a tolerable safety profile up to 20 mg/ day according to a previous phase 1 study. As part of our ongoing efforts in identifying new CXCR4 antagonists, a CXCL12-derived small cyclic peptide (2), which selectively interacts with CXCR4, was identified.19 Although 2 possesses an interesting in vitro and in vivo pharmacological profile, it suffers from degradation in biological fluids (see “Design” paragraph) and its IC50 is in the low micromolar range. Herein, to enhance both the CXCR4 affinity and metabolic stability of 2, we embarked on a lead optimization campaign. By modifying both the N- and C-termini and through a D-amino acid scan, a new, potent, selective, and plasma stable CXCR4 antagonist was identified (10). The novel cyclic peptides were assayed against CXCR4 by measuring their ability to inhibit receptor binding of PE-conjugated-12G5 anti-CXCR4 antibodies and CXCL12-dependent migration in CEM human lymphoblastoid cells, highly expressing CXCR4. Further biological investigations on two different human colon cancer cell lines (HT29 and HCT116) overexpressing CXCR4 confirmed the in vitro efficacy and thus the capability to inhibit the CXCR4 receptor as demonstrated through impairment of CXCL12-mediated cell migration and ERK phosphorylation. A favorable selectivity profile toward CXCR4 over CXCR3 and CXCR7 was also found. Finally, the molecular bases for the improved CXCR4 affinity of 10 were unraveled by NMR and molecular modeling. The improved properties of 10, with respect to 2, might pave the way for the future preclinical development of our peptide in CXCR4 overexpressing colon cancer.

Table 1. Half-Maximal Inhibitory Concentration (IC50, μM, Mean ± SD) of CXCR4 Antagonist Peptides Necessary to Reduce by 50% The Binding to CXCR4 of CXCR4-Specific mAb 12G5 in CCRF-CEM Cellsa compd

sequence

IC50 (μM)

2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 1

H-Arg-Ala-[Cys-Arg-Phe-Phe-Cys]-COOH Ac-Arg-Ala-[Cys-Arg-Phe-Phe-Cys]-COOH Ac-Arg-Ala-[Cys-Arg-Phe-Phe-Cys]-CONH2 H-Arg-Ala-[Cys-Arg-Phe-Phe-Cys]-CONH2 Ac-Arg-Ala-[Cys-Arg-Phe-Phe-DCys]-COOH Ac-Arg-Ala-[Cys-Arg-Phe-DPhe-Cys]-COOH Ac-Arg-Ala-[Cys-Arg-DPhe-Phe-Cys]-COOH Ac-Arg-Ala-[Cys-DArg-Phe-Phe-Cys]-COOH Ac-Arg-Ala-[DCys-Arg-Phe-Phe-Cys]-COOH Ac-Arg-DAla-[Cys-Arg-Phe-Phe-Cys]-COOH Ac-DArg-Ala-[Cys-Arg-Phe-Phe-Cys]-COOH Ac-Ala-Ala-[DCys-Arg-Phe-Phe-Cys]-COOH Ac-Arg-Ala-[DCys-Ala-Phe-Phe-Cys]-COOH Ac-Arg-Ala-[DCys-Arg-Ala-Phe-Cys]-COOH Ac-Arg-Ala-[DCys-Arg-Phe-Ala-Cys]-COOH

6.2 ± 1.3 0.131 ± 0.0014 1.23 ± 0.28 5.9 ± 0.64 5.9 ± 0.42 >10 >10 >10 0.053 ± 0.004 >10 >10 >10 >10 >10 >10 0.006 ± 0.004

a

Each experiment was performed in triplicate.

improved to achieve an optimal in vivo activity. Interestingly, from a structural point of view, 2 shares common pharmacophoric features such as two aromatic and two positively charged residues, with FC131, a cyclic pentapeptide previously described as a potent CXCR4 antagonist.20 Thus, we envisaged that the micromolar activity of 1 may be ascribable to a wrong orientation of those pharmacophoric amino acids within the receptor binding site. To enhance both the peptide CXCR4 affinity and metabolic stability, a systematic modification scheme was designed. Specifically, to increase the stability of 2, the N-terminus was acetylated (3 and 4, Table 1) while the C-terminus was amidated (4 and 5, Table 1). The obtained compounds were validated for inhibiting CXCR4 function through in vitro assays evaluating the anti-CXCR4 antibody (12G5) competition and the CXCL12 dependent migration in CEM human lymphoblastoid cells, highly expressing CXCR4. Next, the most promising peptide emerging from this first set of modifications (3) was subjected to a Damino acid scan (Table 1, 6−12) to investigate the role of each residue’s chirality in the peptide−receptor interaction. A new, stable, and potent CXCR4 analogue (10) was discovered and the specific contribution of each putative pharmacophoric side chains was then investigated by performing an L-Ala scan (Table 1, 13−16). 8370

DOI: 10.1021/acs.jmedchem.6b00695 J. Med. Chem. 2016, 59, 8369−8380

Journal of Medicinal Chemistry

Article

Chemistry. All the peptides were synthesized adopting a classical Fmoc/Boc solid phase peptide strategy. AM-PS resin functionalized with a RINK-amide linker was used for the CONH 2 -terminal peptides (4,5, Table 1), while 2-Cl chlorotrityl resin was selected as solid support for the COOH-terminal peptides (2, 3, 6−16, Table 1) in order to minimize the racemization of the first cysteine residue loaded onto resin.21 The linear eptapetides were build using standard N-Fmoc/tBu solid-phase peptide synthesis strategy. After the cleavage of the linear eptapeptides from the resin, the resulting free thiol groups were oxidized using Iodine to form a disulfide bridge. The synthetic approach adopted for the generation of the peptides enabled the rapid and efficient synthesis of a small library of peptides. CXCR4 Binding Assays. The novel cyclic peptides were assayed against CXCR4 measuring their ability to inhibit receptor binding of PE-conjugated-12G5 anti-CXCR4 antibodies in CEM-CCRF human T leukemia CXCR4 expressing cells.22 As shown in Table 1, the conversion of COOH- into CONH2-termini, in association or not with acetylation of the N-terminus, minimally affected the peptide efficacy. Indeed, 4 and 5 showed IC50 of 1.23 and 5.9 μM, respectively. Conversely, 3, which is the acetyl-derivative of 2, reduced the association of the 12G5 CXCR4-antibody to the receptor with an IC50 of 0.131 μM, thus being about 50-fold more potent than the lead (IC50 = 6.2 μM). Thus, the Nα-acetylation of the Arg1 residue represented a successful modification only in the presence of the COOH-terminus, which appeared to be a critical group for receptor binding. Next, the overall conformation of 3 was modified by performing a D-amino acid scan study. The sequential replacement of each residue with its corresponding D-isomer resulted in analogues (6−12), which mostly had significantly weaker binding affinity than 3, with the lone exception of 10. The latter bears a D-Cys instead of L-Cys at the position 3 and showed an IC50 of 0.053 μM that was 120 times lower than that of 2. In line with our hypothesis, these data clearly pointed out that the conformational changes induced in 10 through the use of a D-amino acid were able to better orient the pharmacophoric side chains to increase the binding. An alanine scan study, restricted to the phenylalanine and arginine residues, was then carried out on 10 to assess their role in the peptide binding. As expected, no or very low CXCR4 binding was observed up to 10 μM concentration of the alanine substituted peptides 13−16. CXCR3 and CXCR7 Binding Assays. The exclusive binding of a drug to its target is required to avoid potential side effects resulting from the interactions with other receptors. Here, we assessed the selectivity profile of our most potent peptide (10) toward other chemokine receptors, specifically CXCR3 and CXCR7. In fact, we evaluated the anti-CXCR3 and anti-CXCR7 efficacy of 10 on a colon cancer cell line (COLO205) overexpressing CXCR323 and a breast cancer cell line (MCF-7) overexpressing CXCR7.24 As shown in Figure 1, 10 did not bind either CXCR7 or CXCR3 up to 10 μM concentration. Interestingly, similar results had been previously obtained for 2,19 indicating that the chemical modifications here introduced have not affected the selectivity profile of the lead. Cell Migration and p-ERK Induction Experiments. CXCL12 dependent migration and p-ERK induction represent major activities elicited by the CXCL12-induced CXCR4 signaling. For these reasons, we evaluated the ability of the

Figure 1. Binding assays on CXCR3 and CXCR7: indirect binding experiments with the anti-CXCR7, clone 11G8, and anti-CXCR3 clone 49801 were conducted in MCF-7 and COLO205, CXCR7, and CXCR3 overexpressing cells, respectively. Data are presented as bar graph showing mean ± SD.

new analogue 10 to inhibit these two CXCR4 functions in comparison with the parent peptide 1. CXCL12 dependent migration of CEM cells was evaluated in the presence of 10 or 2. CEM cells were allowed to migrate toward CXCL12 (100 ng/mL) for 18 h. As shown in Figure 2 and Supporting Information, Figure S1a, 10 inhibited CEM migration more efficiently than 2 and in a dose-dependent manner (Supporting Information, Figure S2). Subsequently, the effect of 2 and 10 on the CXCL12-dependent p-ERK induction (Figure 3) was evaluated, showing that 10 can efficiently inhibit the CXCL12induced p-ERK more efficiently than 2 (Figure 3a−c). Evaluation of 2 and 10 on Human Colon Cancer Cells. To evaluate the CXCR4 antagonistic activity of 10 in a model suitable for future clinical development, we assessed its efficacy on the CXCR4 overexpressing HT29 human colon cancer cell line. As shown in Figure 4, in HT29 cells, 10 binds CXCR4 in a concentration-dependent manner, showing higher potency than 2 (IC50 decreased from 5.2 ± 2.6 μM (2) to 0.064 ± 0.024 μM (10)). Subsequently, the ability of our peptides to inhibit CXCL12 dependent HT29 cells migration was evaluated. Results of these experiments indicated that both 2 and 10 can affect cells migration toward CXCL12, with the latter being effective even at nanomolar concentration (Figure 4b and Supporting Information, Figure S1b). CXCR4 Internalization Assay. To better understand the effect of 2 and 10 on CXCR4 expression and activation, a receptor internalization assay was carried out on the colon 8371

DOI: 10.1021/acs.jmedchem.6b00695 J. Med. Chem. 2016, 59, 8369−8380

Journal of Medicinal Chemistry

Article

Figure 2. Inhibition of CEM cells migration experiments. Migrated cells on the lower surface were fixed, stained with H&E, and counted microscopically. The cells were counted in 10 different consecutive high power fields (magnification 200×). The 0.70% of total input cells migrated in buffer alone containing 1% BSA. The results are expressed as migration index (CXCL12 dependent migration/BSA dependent migration). 2 and 10 were compared to the CXCR4 reference antagonist, 1. Data are presented as bar graph showing mean ± SD.

internalized as indicated by the reduction of membrane fluorescence intensity and formation of CXCR4 containing intracellular vesicle. In the presence of 2 or 10, receptor internalization is impaired as suggested by intracellular vesicles reduction and consistent membrane fluorescence. A slight further reduction in the vesicle size is detected when the cells are treated with 10. Plasma Stability Assay. The effects of 2 modifications on metabolic stability were assessed by incubating 2, 3, and 10 in diluted human plasma at 37 °C adopting a protocol described in the literature.25 Time course aliquots were collected, treated with acetonitrile to precipitate the plasma proteins, and analyzed by ESI-RP-HPLC. As shown in Figure 6, after 60 min, the concentration of 2 was reduced by 50%, resulting in a cleaved form of the peptide lacking the Arg1 residue and the Arg1−Ala2 dipeptide at the N-terminal region (Figure 6b). After 120 min, 2 was fully converted into the cyclic pentapeptide metabolite. Conversely, 3 and 10, both featuring an N-terminal acetyl cap, were stable up to 180 min. Taken together, these results clearly pointed out that the acetyl group at the Nterminal region was sufficient to protect the peptide against human plasma proteolytic degradation. NMR Spectroscopy. NMR experiments were performed on 2 and 10 in order to understand the conformational differences induced in our peptides by the diverse configuration of the Cys3 residue and by the introduction of an acetyl N-cap. Complete 1 H NMR chemical shift assignments (Supporting Information, Tables S1 and S2) were performed for 2 and 10 according to the Wüthrich26 procedure. DQF-COSY,27 TOCSY,28 and NOESY29 experiments, with the support of the XEASY software package,30 were carried out in 200 mM SDS micellar solution. The employment of SDS micelles to investigate the conformational properties is justified on the basis of their interaction with a membrane receptor. For peptides that bind membrane receptors, such as GPCR, the use of membrane mimetic solution is suggested, assuming a membrane-assisted mechanism of interactions between the peptides and their receptors,31 hence micelle solutions have been extensively used for peptide hormone conformational studies.32 For both

Figure 3. CXCL12-mediated p-ERK induction assay: CEM cells were serum-starved and treated with CXCL12 (100 ng/mL) in the presence of 2 and 10.

cancer cell line HCT116 engineered to express human GFPtagged CXCR4 receptor. HCT116-GFP-CXCR4 cells were treated with 2 and 10, then stimulated with 50 ng/mL CXCL12 to induce receptor internalization and observed by confocal microscopy. As showed in Figure 5, in the absence of CXCL12, CXCR4 localizes at both the membrane and cytoplasmic level, with some accumulation at the perinuclear region. After 45 min of stimulation with CXCL12, GFP-CXCR4 is extensively 8372

DOI: 10.1021/acs.jmedchem.6b00695 J. Med. Chem. 2016, 59, 8369−8380

Journal of Medicinal Chemistry

Article

Figure 4. (a) CXCR4 binding experiments on 2 and 10 in human colon cancer cells. CXCR4 binding was indirectly evaluated through inhibition of 12G5 anti-CXCR4 antibody binding in HT29 human colon cancer CXCR4 expressing cells. (b) CXCL12-induced migration assays on human colon cancer cells HT29. 0.81% of total input cells migrated in buffer alone containing 1% BSA.

peptides, NMR parameters indicated the presence of β-turn structures. In particular, NOE contacts between Hα-NHi+2 of Arg1 and D-Cys3 and Arg4 and Phe6 suggested turn structures encompassing residues 1−3 and 3−6, respectively. The presence of turn structures is confirmed by low values of temperature coefficients of amide protons (|Δδ/ΔT| < 4 ppb/ K) of Phe5 and Phe6 for both peptides and D-Cys3 for 10 (Supporting Information, Tables S1 and S2). Upfield shifts of side chain protons of the Arg4 and Phe5 residues and NOE contacts between Arg4 and Phe5 and between Phe5 and Phe6 point to a spatial proximity of these couple of side chains in both peptides. For both peptides, NMR constraints from SDS micelle solution were used as the input data for a simulated annealing structure calculation. Two ensembles of well-defined structures were obtained. In fact, the 10 lowest energy structures for 2 (Figure 7a) and 10 (Figure 7b) showed a backbone RMSD of 0.19 and 0.17 Å, respectively, that satisfied the NMR-derived constraints (violations smaller than 0.40 Å). Predicted turn structures were found in both peptides: a γ-turn along residues 1−3 and a type IV β-turn along residues 3−6. Interestingly, in 2 calculated structures, Cys3 shows a positive value of the dihedral angle φ (∼60°), which could explain the observed constructive replacement with a D-Cys residue. The main difference between the two peptide structures consists in the different spatial disposition of the Arg1 side chain, which is close to Phe5 and Phe6 only in 10 as confirmed by various NOE

interactions between Arg1 with Phe5 and acetyl protons with Phe6 observable only in the NOESY spectrum of 10 while missing in that of 2. Geometry of D-Cys residue and lipophilicity of the acetyl moiety in 10 must be responsible for the observed difference. Interestingly, Ala- and D-scan results clearly demonstrated that both functionality and spatial arrangement of Arg1 side chain matter for the activity of our derivatives. Hence, Arg1 side chain reorientation in 10 is likely to contribute to the increased potency of the peptide. This issue has been further investigated by molecular modeling studies. Molecular Modeling. To investigate at an atomic level the binding to CXCR4 of the most potent peptide of the series, namely 10, extensive computational studies were performed. In the present case, straightforward docking methods did not represent the best option, as they cannot thoroughly sample the large conformational space of mid−high molecular weight peptides upon binding. For this reason, and to take into account the receptor flexibility and the biological environment, we performed an over 100 ns molecular dynamics (MD) simulation on the 10/CXCR4 complex in explicit solvent and membrane. The starting binding conformation was obtained through a manual docking of 10 in the crystal structure of CXCR4 in complex with the 16-mer cyclic peptide CVX15 (PDB code 3OE0). Specifically, a superposition between 10 and CVX15 was accomplished based on previous structure− activity relationship (SAR) studies on potent CXCR4 peptide 8373

DOI: 10.1021/acs.jmedchem.6b00695 J. Med. Chem. 2016, 59, 8369−8380

Journal of Medicinal Chemistry

Article

Figure 7. Superposition of the 10 lowest energy conformers of 2 (a) and 10 (b). Structure models were superimposed using the backbone heavy atoms. Heavy atoms have different colors (carbon, orange/ green; nitrogen, blue; oxygen, red; sulfur, yellow). Hydrogen atoms are hidden for clarity reasons. (c) Superposition of 2 (green) with 10 (orange) using the three main pharmacophoric points: the guanidino group of the Arg4 and the centroids of both the Phe5 and Phe6 phenyl rings.

Figure 5. CXCR4 internalization assay: HCT116 colon cancer cells were engineered to express GFP-tagged CXCR4; α-tubulin staining was used to define cells cytoplasm and shape. Cells were treated for 45 min with 100 nM of either 2 or 10 and then for 45 min with CXCL12.

antagonists, which share one positively charged and two aromatic side chains as common pharmacophoric features.33 (see Experimental Section for details). At the very beginning of the MD calculations, 10 slightly rearranges at the CXCR4 binding site, assuming a conformation that is almost conserved for the rest of the simulation (Supporting Information, Figure S3). In this pose (Figure 8), the peptide Arg1 and Arg4 side chains make tight salt bridges with the D187 and D97 carboxylate groups, respectively, whereas the Phe5 and Phe6 side chains are buried into a semiaromatic pocket where they can establish favorable interactions with the side chains of Y116, L120, R188, F199, H203, Y255, and I259. Additionally, water-mediated hydrogen bonds are formed between the Arg1 and N37 side chains and between the Phe6 backbone amide and the E288 carboxylate moiety. Interestingly, a number of intramolecular interactions can be also observed, which are likely to stabilize the peptide bioactive conformation. In particular, the Arg1 side chain is hydrogen bonded to the Cys7 terminal carboxylate and to the N-acetyl carbonyl group, while an additional H-bond is established between the D-Cys3 CO and the Cys7 NH groups. Overall, this interaction pattern is consistent with the low nanomolar IC50 of 10 and in agreement with mutagenesis data reporting residues such D97, H113, Y116, D187, and E288 as critical for ligand binding to CXCR4.33,34 Furthermore, comparative studies on 2 (see the Supporting Information for details) helped us to elucidate the molecular bases for the improved CXCR4 binding affinity of 10 with respect to the lead peptide. In fact, some differences can be observed in the binding mode of the two peptides, which are ascribable to the inverted chirality of the Cys3 residue and the absence of the Nacetyl cap in 2 (Supporting Information, Figure S4). For instance, differently from 10, the Arg1 side chain of 2 cannot contact D187 side chain but extends out of the binding site

Figure 6. (a) Human plasma stability profiles of 2 and 10 after different intervals of incubation with human plasma. Relative concentrations of peptides were determined by integration of the A230 peaks from RP-HPLC (b) ESI-MS characterizations of 2 and 10 at different intervals of incubation.

8374

DOI: 10.1021/acs.jmedchem.6b00695 J. Med. Chem. 2016, 59, 8369−8380

Journal of Medicinal Chemistry

Article

Figure 8. (a) Side view of the predicted CXCR4/10 complex embedded in the POPC bilayer. CXCR4 is represented as gray cartoons, while the peptide is shown as orange spheres. Lipids are shown as yellow sticks. (b) Binding mode of 10 (orange sticks) at the CXCR4 receptor (gray cartoons) obtained through over 100 ns long MD simulations. Receptor amino acids and waters important for peptide binding are shown as sticks. Hydrogen bonds are displayed as dashed black lines. Nonpolar hydrogens are omitted for clarity reasons.

impairs CXCR4 internalization in human colon cancer cells overexpressing CXCR4, which might lay the foundations for a future preclinical investigation of our novel peptide. In this perspective, in cell NMR studies are now in progress in our laboratories to ascertain the competition mechanism between 10 and 1 in order to get more insights into the pharmacodynamics properties of our peptides. Nonetheless, the modeling data herein reported pave the way for a further optimization of the 10 IC50 and another round of synthesis is ongoing accordingly. Finally, the high stability of 10 in human plasma is surely an encouraging starting point for more extensive in vivo pharmacokinetic studies. In this light, a GPCR gene expression profiling could be particularly important in the evaluation of possible toxicities beyond those arising from CXCR4 blockade.

interacting with the lipids polar heads. Moreover, the intramolecular H-bonds formerly described for 10 (i.e., between the Arg1 side chain and the Cys7 terminal carboxylate) are not observed in the case of 2, turning out in a lower conformation stability that might negatively impact the binding. The interaction models presented here also allow rationalizing the effects of all the introduced chemical modifications on our peptides potency. In particular: (i) mutation into Ala of Arg1, Arg4, Phe5, and Phe6 (13−16) is detrimental for peptide affinity toward CXCR4 because the side chains of these residues represent the major anchor points for receptor binding; (ii) the chiralities of the amino acids occupying positions from 3 to 7 (6−12), particularly D-Cys3 and Cys7, are crucial to stabilize the binding of the peptide pharmacophoric groups at the CXCR4 receptor; (iii) amidation of the Cys7 terminal carboxylate and/or removal of the N-Ac cap (2−5) lower the peptide affinity toward CXCR4 because such groups participate in intramolecular interactions that contribute to stabilize the peptide binding conformation. Finally, our model suggests that at least one of the two Phe residues in 10 might be substituted with bulkier aromatic residues, such as 1- or 2Nal, providing some hints for a further optimization of our peptides potency.



EXPERIMENTAL SECTION

Chemistry. Materials. Nα-Fmoc-protected amino acids, 2-Cl-trtCl resin, Fmoc-Rink amide-Am resin, O-benzotriazole-N,N,N′,N′-tetramethyl-uroniumhexafluorophosphate (HBTU), N,N-diisopropylethylamine (DIEA), triisopropylsilane (TIS), trifluoroacetic acid (TFA), and piperidine were purchased from IRIS Biotech (Marktredwitz, Germany), N-hydroxybenzotriazole (HOBt), N,N-dimethylformamide (DMF), dichloromethane (DCM), and from Sigma-Aldrich (Milano, Italy). Peptide synthesis solvents, reagents, H2O, and CH3CN for HPLC were reagent grade and were acquired from commercial sources (Sigma-Aldrich, Milano, Italy) and used as received unless otherwise noted. Peptides were purified by preparative HPLC (Shimadzu HPLC system) equipped with a C18-bounded preparative RP-HPLC column (Phenomenex Kinetex 21.2 mm × 150 mm 5 μm). Peptides were analyzed by analytical HPLC (Shimadzu Prominance HPLC system) equipped with a C18-bounded analytical RP-HPLC column (Phenomenex Kinetex, 4.6 mm × 250 mm 5 μM) using a gradient elution (10−90% acetonitrile in water (0.1% TFA) over 15 min; flow rate = 1.0 mL/min; diode array UV detector). Molecular weights of compounds were confirmed by ESI-mass spectrometry using an Agilent 6110 quadrupole LC/MS system. General Procedure for COOH-Terminal Peptides. 2-Cl-trtCl resin (62.0 mg, 1.60 mmol/g) was swollen in DCM over 0.5 h, and a solution of Fmoc-L-Cys(trt)-OH (87.9 mg, 0.15 mmol, 1.5 equiv) and DIPEA (26 μL, 1.5 equiv) in DCM (2 mL) was added. The mixture was stirred for 10 min. An additional amount of DIPEA (34.8 μL, 0.2 mmol, 2 equiv) was added, and the mixture was then shaken for 1 h. The residual chloride groups contained in the resin were capped by adding MeOH (200 μL) in DCM (2 mL) and stirring for 15 min to avoid eventually parallel synthesis of side products. Fmoc group removal was performed using 20% piperidine in DMF (1 × 5 min and 1 × 25 min). The peptide resin was then washed with DCM (3 × 0.5 min) and DMF (3 × 0.5 min), and positive Kaiser ninhydrin35 and



CONCLUSION In recent years, the CXCL12/CXCR4 pathway has emerged as a major target for the therapy of multiple diseases including various types of blood and solid tumors. For this reason, a number of CXCR4 antagonists are now used or are entering as diagnostic and therapeutic tools for cancer treatment. However, in spite of the medicinal chemists’ efforts, the majority of the available CXCR4 antagonists suffer pharmacokinetic or toxicity issues, prompting the design of new anti-CXCR4 ligands with improved pharmacological properties. In this scenario, some of us recently developed a CXCL12-derived cyclic peptide as selective CXCR4 antagonist (2). Herein, through a rational lead optimization strategy, we improved both the binding potency which was in the low micromolar range and the metabolic stability of 2, thus obtaining a selective and plasma stable CXCR4 antagonist (10). Ala-amino acid scan, extensive NMR, and computational studies elucidated the binding mode of 10 to CXCR4, providing reasons for the increased binding with respect to 2. In parallel, cell-based assays revealed that 10 inhibits both the CXCL12-induced migration and ERK phosphorilation, which are hallmarks for monitoring CXCR4 functions. Furthermore, we were able to demonstrate that 10 8375

DOI: 10.1021/acs.jmedchem.6b00695 J. Med. Chem. 2016, 59, 8369−8380

Journal of Medicinal Chemistry

Article

TNBS36 tests were observed. Fmoc-L-Phe-OH (155.0 mg, 0.4 mmol, 4 equiv) or Fmoc-D-Phe-OH (155.0 mg, 0.4 mmol, 4 equiv) or Fmoc-LAla-OH (124.5 mg, 0.4 mmol, 4 equiv), Fmoc-L-Arg(Pbf)-OH (259.5 mg, 0.4 mmol, 4 equiv) or Fmoc-D-Arg(Pbf)-OH (259.5 mg, 0.4 mmol, 4 equiv) or Fmoc-L-Ala-OH (124.5 mg, 0.4 mmol, 4 equiv), Fmoc-L-Cys(Trt)-OH (234.3 mg, 0.4 mmol, 4 equiv) or Fmoc-DCys(Trt)-OH (234.3 mg, 0.4 mmol, 4 equiv), Fmoc-L-Ala-OH (124.5 mg, 0.4 mmol, 4 equiv) or Fmoc-D-Ala-OH (124.5 mg, 0.4 mmol, 4 equiv) were sequentially added sequentially to the resin bound H-LCys(Trt). Each coupling reaction was achieved using a 4-fold excess of amino acid with HBTU (151.7 mg, 0.4 mmol, 4 equiv) and HOBt (61.2 mg, 0.4 mmol, 4 equiv) in the presence of DIPEA (140 μL, 0.8 mmol, 8 equiv) in DMF. Fmoc deprotections were accomplished with 20% piperidine in DMF solution (1 × 5 min, 1 × 25 min). Washings with DMF (3 × 0.5 min) and DCM (3 × 0.5 min) were performed through every coupling/deprotection step. Kaiser ninhydrin and TNBS tests were employed for monitoring the progress of peptide synthesis. For the peptides bearing an acetyl group at the N-terminus, after removing the last Fmoc group, the resin bound peptide was treated with Ac2O (19 μL, 0.2 mmol, 2 equiv) and DIPEA (35 μL, 0.2 mmol, 2 equiv) in DCM (2 mL) and the mixture was shaken for 1 h. A negative Kaiser ninhydrin and TNBS tests were observed. General Procedure for CONH2-Terminal Peptides. Rink amide resin (208 mg, 0.48 mmol/g) was swollen in DMF over 0.5 h, and the Fmoc protecting group of the linker was removed with 20% piperidine in DMF solution (1 × 5 min and 1 × 25 min). The resin was washed with DMF (3 × 0.5 min) and DCM (3 × 0.5 min) and positive Kaiser ninhydrin and TNBS tests were observed. The linear sequences were assembled sequentially by adding Fmoc-L-Cys(trt)-OH (234.3 mg, 0.4 mmol, 4 equiv), Fmoc-L-Phe-OH (154.98 mg, 0.4 mmol, 4 equiv), Fmoc-L-Arg(Pbf)-OH (259. 51 mg, 0.4 mmol, 4 equiv), Fmoc-LCys(trt)-OH (234.3 mg, 0.4 mmol, 4 equiv), and Fmoc-L-Ala-OH (124.54 mg, 0.4 mmol, 4 equiv). Each coupling reaction was achieved using a 4-fold excess of amino acid in the presence of HBTU (151.7 mg, 0.4 mmol, 4 equiv), HOBt (61.2 mg, 0.4 mmol, 4 equiv), and DIPEA (140 μL, 0.8 mmol, 8 equiv) in DMF (3 mL). Fmoc deprotections were accomplished with 20% piperidine solution in DMF (1 × 5 min, 1 × 25 min). Washings with DMF (3 × 0.5 min) and DCM (3 × 0.5 min) were performed through every coupling/deprotection step. Kaiser ninhydrin and TNBS tests were employed for monitoring the progress of peptide synthesis. For the peptides bearing an acetyl group at the N-terminus, after removing the last Fmoc group, the resin bound peptide was treated with Ac2O (19 μL, 0.2 mmol, 2 equiv) and DIPEA (35 μL, 0.2 mmol, 2 equiv) in DCM (2 mL) and the mixture was shaken for 1 h. Negative Kaiser ninhydrin and TNBS tests were observed. General Procedure for Peptide Oxidation and Purification. The peptide was released from the solid support and all the protecting groups cleaved, treating the resin with TFA/DCM/TIS (80/15/5, v/ v/v) (3 mL solvent/0.1 mmol) for 2 h. The resin was then filtered off, and the crude linear peptide was recovered by precipitation with chilled ether to give a powder. The crude peptide (0.1 mmol) was dissolved in 30 mL of glacial AcOH, and 200 μL of HCl 1N were added to improve the solubility of the peptide. Next, a solution of iodine (40 mg, mmol) in glacial AcOH (5 mL) was added and the mixture was mechanical stirring for 12 h at room temperature. The resulting mixture was concentrated in vacuo by rotatory evaporator (1/4 of the original volume) and transferred to a 50 mL conical centrifuge tube, where the oxidated crude peptide was precipitated by adding 40 mL of chilled ether, recovered by filtration, and dried overnight. Final peptide purification was achieved by preparative RPHPLC in 0.1% TFA with an ACN gradient (10−90% ACN in H2O over 15 min, flow rate of 15 mL/min) on a Phenomenex Kinetex C18 column (21.2 mm × 150 mm 5 μm). Analytical RP-HPLC were performed in 0.1% TFA with an ACN gradient (10−90% ACN in H2O over 20 min, flow rate of 1.0 mL/min) on a Phenomenex Kinetex C18 column (0.46 mm × 150 mm 5 μm). 2: Purity >95%, tR 12.2 min (analytical HPLC, 10−90% ACN in H2O (0.1% TFA) over 15 min, flow rate of 1.0 mL/min); molecular

formula, C39H57N13O8S2; calculated mass, 899.4; found, 900.3 (M + H+), 450.7 (M + 2H+)/2. 3: Purity >95%, tR 12.5 min (analytical HPLC, 10−90% ACN in H2O (0.1% TFA) over 15 min, flow rate of 1.0 mL/min); molecular formula, C41H59N13O9S2; calculated mass, 941.4; found, 942.6 (M + H+), 471.4 (M + 2H+)/2. 4: Purity >95%, tR 12.0 min (analytical HPLC, 10−90% ACN in H2O (0.1% TFA) over 15 min, flow rate of 1.0 mL/min); molecular formula, C39H57N13O8S2; calculated mass, 898.4; found, 899.6 (M + H+), 450.0 (M + 2H+)/2. 5: Purity >95%, tR 12.2 min (analytical HPLC, 10−90% ACN in H2O (0.1% TFA) over 15 min, flow rate of 1.0 mL/min); molecular formula, C41H60N14O8S2; calculated mass, 940.4; found, 941.6 (M + H+), 471.0 (M + 2H+)/2. 6: Purity >95%, tR 12.6 min (analytical HPLC, 10−90% ACN in H2O (0.1% TFA) over 15 min, flow rate of 1.0 mL/min); molecular formula, C41H59N13O9S2; calculated mass, 941.4; found, 942.5 (M + H+), 471.8 (M + 2H+)/2. 7: Purity >95%, tR 12.6 min (analytical HPLC, 10−90% ACN in H2O (0.1% TFA) over 15 min, flow rate of 1.0 mL/min); molecular formula, C41H59N13O9S2; calculated mass, 941.4; found, 942.4 (M + H+), 471.8 (M + 2H+)/2, 964.5 (M + Na+). 8: Purity >95%, tR 12.4 min (analytical HPLC, 10−90% ACN in H2O (0.1% TFA) over 15 min, flow rate of 1.0 mL/min); molecular formula, C41H59N13O9S2; calculated mass, 941.4; found, 942.4 (M + H+), 471.8 (M + 2H+)/2. 9: Purity >95%, tR 12.4 min (analytical HPLC, 10−90% ACN in H2O (0.1% TFA) over 15 min, flow rate of 1.0 mL/min); molecular formula, C41H59N13O9S2; calculated mass, 941.4; found, 942.4 (M + H+), 471.8 (M + 2H+)/2. 10: Purity >95%, tR 12.6 min (analytical HPLC, 10−90% ACN in H2O (0.1% TFA) over 15 min, flow rate of 1.0 mL/min); molecular formula, C41H59N13O9S2; calculated mass, 941.4; found, 942.4 (M + H+), 471.8 (M + 2H+)/2, 964.4 (M + Na+). 11: Purity >95%, tR 12.5 min (analytical HPLC, 10−90% ACN in H2O (0.1% TFA) over 15 min, flow rate of 1.0 mL/min); molecular formula, C41H59N13O9S2; calculated mass, 941.4; found, 942.4 (M + H+), 471.8 (M + 2H+)/2. 12: Purity >95%, tR 12.5 min (analytical HPLC, 10−90% ACN in H2O (0.1% TFA) over 15 min, flow rate of 1.0 mL/min); molecular formula, C41H59N13O9S2; calculated mass, 941.4; found, 942.6 (M + H+), 471.8 (M + 2H+)/2, 964.5 (M + Na+). 13: Purity >95%, tR 15.1 min (analytical HPLC, 10−90% ACN in H2O (0.1% TFA) over 15 min, flow rate of 1.0 mL/min); molecular formula, C38H52N10O9S2; calculated mass, 856.3; found, 857.4 (M + H+). 14: Purity >95%, tR 15.1 min (analytical HPLC, 10−90% ACN in H2O (0.1% TFA) over 15 min, flow rate of 1.0 mL/min); molecular formula, C38H52N10O9S2; calculated mass, 856.3; found, 857.4 (M + H+). 15: Purity >95%, tR 13.6 min (analytical HPLC, 10−90% ACN in H2O (0.1% TFA) over 15 min, flow rate of 1.0 mL/min); molecular formula, C35H55N13O9S2; calculated mass, 865.4; found, 866.4 (M + H+), 433.7 (M + 2H+). 16: Purity >95%, tR 13.8 min (analytical HPLC, 10−90% ACN in H2O (0.1% TFA) over 15 min, flow rate of 1.0 mL/min); molecular formula, C35H55N13O9S2; calculated mass, 865.4; found, 866.4 (M + H+), 433.7 (M + 2H+). Human Plasma Stability Assay. Human plasma, H2O, ACN, and TFA were obtained from Sigma-Aldrich (Milano, Italy) and used without further purification. Human plasma was diluted to 50% in pure water. Analytical HPLC-ESI-MS was performed on an Agilent Technologies 1200 series equipped with an Agilent Technologies 6110 quadrupole LC/MS using a Phenomenex Luna C18 column (5 μm, 4.6 mm × 150 mm) and H2O (0.1% v/v TFA)/ACN (0.1% v/v TFA) as eluents. A solution of 1 mg/mL of each peptide was prepared in water, and 150 μL aliquots were mixed with 150 μL of prewarmed (37 °C) plasma. At selected time points (0, 30, 60, 120, and 180 min), samples (50 μL) were collected and mixed with 1% TFA in acetonitrile (75 8376

DOI: 10.1021/acs.jmedchem.6b00695 J. Med. Chem. 2016, 59, 8369−8380

Journal of Medicinal Chemistry

Article

μL) to precipitate plasma proteins which were deleted by centrifugation at 13000 rpm for 10 min. The supernatant was analyzed by HPLC-ESI-MS using a Phenomenex Luna C18 column (5 μm, 4.6 mm × 150 mm) and an elution gradient of 10−90% solvent B over 20 min (solvent A, 0.1% TFA in water; solvent B, 0.1% TFA ACN) at a flow rate of 1 mL/min. Biological Studies. Binding Assay. CXCR4 binding was evaluated as previously described.22 Briefly, 5 × 105 CCRF−CEM, HT29 cells were preincubated with increasing peptide concentrations (0.01 μM, 0.1 μM, 1 μM, 10 μM) in the binding buffer (PBS 1× plus 0.2% BSA and 0.1% NaN3) for 30 min at 37 °C, 5% CO2 and then labeled for 30 min using anti-CXCR4 PE-antibody (FAB170P, clone 12G5, R&D Systems, Minneapolis, MN, USA). To evaluate the specific peptide binding to CXCR4, the experiments were also conducted in COLO205, human colon cancer cells, overexpressing CXCR3, using anti-CXCR3 FITC-antibody (R&D FAB160F clone 49801) and in MCF-7, human breast cancer cell line, overexpressing CXCR7, using anti-CXCR7 APC-antibody (R&D FAB4227A clone 11G8). The cells were counted through a FACS Canto II cytofluorometer (Becton Dickinson Immunocytometry Systems, Mountain View, CA, USA). Migration Assay. Migration was assayed in 24-well Transwell chambers (Corning Inc., Corning, NY) using inserts with 5 and 8 μm (optimal for lymphocytes and epithelial cells, respectively) pore membranes. Membranes were precoated with collagen (human collagen type I/III) and fibronectin (20 μg/mL each). Cells were placed in the upper chamber (2 × 105cells/well) in culture medium containing 1% BSA (migration media) in the presence of the peptide; 100 ng/mL CXCL12 was added to the lower chamber. After 18 h of incubation, cells atop the filter were removed using a cotton wool swab; migration of cells in the lone medium (control) was compared with that observed in media containing CXCL12. The cells were counted in 10 different fields (original 400× magnification). The migration index was defined as the ratio between migrating cells in the experimental group and migrated cells in the control group. Immunoblotting. Cells were homogenized in lysis buffer (40 mM Hepes pH 7.5, 120 mM NaCl, 5 mM MgCl2, 1 mM EGTA, 0.5 mM EDTA, 1% Triton X-100) containing protease (Complete TabletsEDTA free, Roche) and phosphatase inhibitors (20 mM α-glycerol-3phosphate, 2.5 mM Na-pyrophosphate). The following primary antibodies were used: anti-p-ERK (sc7383, Santa Cruz Biotechnology, Inc., Santa Cruz, CA, USA) and anti-ERK2 (sc 154G, Santa Cruz Biotechnology CA, USA). P-ERK induction was plotted as the ratio between p-ERK in the presence of CXCL12 (100 ng/mL) and p-ERK in free serum. Receptor Internalization Assay. HCT116-GFP-CXCR4 cell line was established in our lab by transfecting pEGFP-CXCR4 plasmid (Origene, USA) into the parental HCT116 cell line; GFP-CXCR4 expression in selected HCT116 clones was then confirmed by flow cytometry and immunofluorescence using anti-CXCR4 12G5 antibody (not shown). Subconfluent HCT116-GFP-CXCR4 cells were grown on 10 mm coverslips in DMEM plus 10% FBS-1% P/S-1% glutammine for 1 day, washed with PBS plus 0.5% BSA, and equilibrated in DMEM plus 0.5% BSA (Sigma-Aldrich). Cells were then exposed to 2 (0.1 μM) or 10 (0.1 μM) for 45 min at 37 °C. Cells were then treated with 50 mg/mL CXCL12 and incubated at 37 °C for 45 min, fixed with 4% formaldehyde for 10 min, washed three times in PBS, permeabilized with 0.1% Triton X-PBS for 10 min, and washed three times again in PBS. Cells were finally stained with αtubulin/Alexa Fluor 594 and DAPI and inspected with a Zeiss LSM 510 confocal microscope equipped with Zeiss 2014 software. NMR Spectroscopy. The samples in micelle solution (200 mM of SDS-d25) for NMR spectroscopy were prepared by dissolving the peptides in 0.18 mL of 1H2O (pH 5.5) and 0.02 mL of 2H2O to obtain a concentration 2 mM of peptides. 2D DQF-COSY,27 TOCSY,28 and NOESY29 spectra were recorded on a Varian INOVA 700 MHz spectrometer equipped with a z-gradient 5 mm triple-resonance probe head at a temperature of 298 K. The water signal was suppressed by gradient echo.37 A mixing time of 80 and 200 ms was used for the TOCSY and NOESY experiments, respectively. The interactive program package XEASY30 was used for the NMR analysis of DQF-

COSY, TOCSY, and NOESY spectra. 3JHN‑Hα coupling constants were acquired from 1D 1H NMR and 2D DQF-COSY spectra. The temperature coefficients of the amide proton chemical shifts were calculated from 1D 1H NMR and 2D TOCSY experiments executed at different temperatures in the range 298−313 K by means of linear regression. Structural Determinations. The NOE cross peaks were integrated with the XEASY program and were converted into upper distance bounds using the CALIBA software provided by the program package DYANA.38 Only NOE derived constraints (Supporting Information, Tables S3 and S4) were considered in the annealing procedures. An ensemble of 100 structures was generated with the simulated annealing calculations followed by successive steps of restrained and unrestrained energy minimization using the Discover algorithm (Accelrys, San Diego, CA). From the produced 100 conformations, 10 structures were chosen, whose interproton distances best fitted NOE derived distances. Molecular Modeling. A manual docking of the 10 averaged NMR structure in the crystal structure of CXCR4 in complex with the 16mer cyclic peptide antagonist CVX15 (PDB code: 3OE0) was first accomplished. Specifically, we superimposed the coordinates of the Cα carbons of Arg4, Phe5, and Phe6 in 10 with the Cα carbons of Arg1, Arg2, and Nal3 in CVX15, which was then removed from the complex. Subsequently, to limit steric clash within the CXCR4 binding cavity, the side chains of Arg1 and Arg4 in 10 were manually extended to interact with Asp181 and Asp97, respectively. Prior to MD simulations, the receptor first N-terminal (K25) and the last Cterminal (A301) residues were capped with ACE and NME, respectively, while missing intracellular loops (ICL1 and ICL3) were added and refined using Prime.39 Histidine tautomeric/protonation states were assigned through the PROPKA module of the Maestro Protein Preparation Wizard.40 The refined complex was embedded in a POPC (1-palmitoyl-2-oleoylphosphatidylcholine) phospholipids bilayer to mimic the physiological environment and then submitted to an over 100 ns MD simulation with NAMD 2.9.41 In detail, a 94 Å × 94 Å (in x and y axes) pre-equilibrated POPC phospholipid bilayer was first created with the aid of the membrane-builder tool of CHARMM-GUI.org (http://www.charmm-gui.org). With the purpose of placing the receptor into the bilayer, a hole was generated and all lipids in close contact (